Predictors for mortality.

The first column gives the crude hazard ratios (HRs) for the different predictors estimated in univariable Cox regression. The second and third column present the adjusted HRs estimated in the multivariable models containing the total and the primary COCI, respectively. Besides of the variables listed in the table, the model with the total COCI considers 41 ATC-codes and 4 ICD-10-codes. The model with the primary COCI considers 39 ATC-codes and 4 ICD-10-codes. See S1 Table for the adjusted HRs of the ATC-codes and ICD-10-codes. Both multivariable models take into account the interaction between age and sex, thus the estimated effect of age differs between the sexes and the effect of sex differs between different age values, as described in the table.